6 New Developments in Intelligence Theory and Assessment
Implications for personnel selection
Charles Scherbaum
Baruch College, City University Of New York, USA
Harold Goldstein
Baruch College, City University Of New York, USA
Rachel Ryan
Baruch College, City University Of New York, USA
Paul Agnello
Baruch College, City University Of New York, USA
Ken Yusko
Marymount University, USA
Paul Hanges
University Of Maryland, USA
DOI: 10.4324/9781315742175-10
Intelligence is an individual difference that is arguably more important than ever for success in the constantly changing and increasingly complex modern business world. Despite its importance and the dramatic changes that have occurred in the nature of work, the conceptualization and use of intelligence in personnel selection has changed very little over the past seventy years (Scherbaum, Goldstein, Yusko, Ryan, & Hanges, 2012). Although the field of personnel selection has only incrementally evolved in its thinking about intelligence, many other fields (e.g., clinical and cognitive psychology, developmental and educational research, and the neurosciences) have been very active in conducting modern intelligence research (Goldstein, Scherbaum, & Yusko, 2009). These fields have made considerable progress in understanding the intelligence construct, its role in the modern world, and how it can be measured. However, the field of personnel selection has yet to take advantage of these developments. As some have argued, the tests that we commonly use have not substantially evolved since their inception (Thorndike, 1997). As a result, a great opportunity to better understand, measure, and use intelligence in personnel selection is being missed.
This chapter reviews some of the innovations and developments in the conceptualization and measurement of intelligence that have the potential to impact personnel selection. Specifically, we review the developments in modern conceptualizations of psychometric approaches to intelligence (e.g., CHC model, XBA), cognitive approaches to intelligence (e.g., neuropsychological approaches, PASS model), and modern intelligence test design principles as well as the implications of these developments for personnel selection.
Intelligence and personnel selection
Researchers have studied intelligence and various aspects of it (e.g., cognitive ability, general mental ability, g) and its impact on a wide array of criteria for over a century (Schmidt & Hunter, 1998). In this research, the psychometric approach based on Spearman’s (1927) work (i.e., psychometric g) has been the dominant way of operationalizing and understanding intelligence in applied psychology (Neisser et al., 1996; Scherbaum et al., 2012). The dominance of this approach is not surprising given that tests measuring this cognitive ability are related to performance in academic and work contexts (Herrnstein & Murray, 1994; Schmidt & Hunter, 1998). However, these tests are also associated with large test score differences between particular racial/ethnic groups (Hough, Oswald, & Ployhart, 2001; Roth, Bevier, Bobko, Switzer, & Tyler, 2001). These group differences in test scores can create disparities on important outcomes such as school admissions and employment decisions.
Despite the dominance of Spearman’s (1927) psychometric approach, researchers are increasingly recognizing the limitations of this approach, including how well it captures the full construct space of intelligence, that the measures based on it may be partly responsible for the observed score differences between groups, and that improved prediction may be possible with modern theoretical conceptualizations and measures (Fagan & Holland, 2007; Goldstein et al., 2009; Nisbett, Aronson, Blair, Dickens, Flynn, Halpern, & Turkheimer, 2012; Scherbaum et al., 2012; van der Maas et al., 2006). In the subsequent sections of this chapter, we describe some of the modern research on intelligence and its measurement that has the potential to address some of these limitations and contribute to our understanding of intelligent behavior in the workplace.
Modern conceptualization of psychometric approaches to intelligence
In the early days of intelligence research, the development of theoretical models of intelligence was a vibrant area of activity. As Wechsler (1975) noted, as far back as 1921 there were as many theories and definitions of intelligence as there were theorists. Over a relatively short period of time, the applied areas of psychology coalesced around the Spearman (1927) model of psychometric g. In Spearman’s theory, there is only a single latent construct (i.e., g) that is needed to account for variation in performance on tests designed to assess cognitive abilities. In the area of personnel selection, the tenets of this model still guide most thinking on the theory and measurement of intelligence (Goldstein et al., 2009).
Contemporary thinking on intelligence in other areas of psychology and research has long postulated that intelligence is a network of different cognitive constructs rather than a single entity (Gottfredson, 2009; Horn & Blankson, 2012; Jensen, 1998; Reeve & Bonaccio, 2011; Schneider & McGrew, 2012). Modern theories have focused on developing hierarchically arranged taxonomies of these abilities. The most supported, accepted, and influential of these models is the Cattell–Horn–Carroll (CHC) model of intelligence (McGrew, 1997; Schneider & McGrew, 2012). The CHC model represents the integration of Carroll’s (1993) three-stratum theory of intelligence with Horn and Cattell’s (1966) theory of fluid and crystalized intelligence.
This theory describes the key dimensions of intelligence at three hierarchical levels of specificity. At the highest and broadest level of this theory is a single general ability (stratum III). The next level (stratum II) includes a number of broad cognitive abilities including fluid reasoning, short-term memory, long-term memory, processing speed, reaction and decision speed, psychomotor speed, comprehension/knowledge (i.e., crystalized intelligence), domain-specific knowledge, reading and writing, quantitative knowledge, visual processing, auditory processing, and three other abilities related to sensory functioning (see Schneider & McGrew, 2012, or Schneider & Newman, in press, for detailed reviews). As Schneider and McGrew (2012) note, the factors at the second level can be organized into abilities related to acquired knowledge, abilities that are independent of a specific domain, and those related to sensory-motor domains. At the lowest and most specific level (stratum I) are sixty-four narrow cognitive abilities (e.g., induction, sequential reasoning, perceptual speed).
The CHC model has served as the theoretical foundation for the revision of many existing tests of intelligence (e.g., Stanford-Binet 5th edition, Woodcock-Johnson III) and the development of some new tests (Keith & Reynolds, 2010). Also, a substantial amount of research has focused on operationalizing this theory into measurement practices. For example, the cross-battery assessment approach (XBA) of Flanagan and colleagues (Flanagan & McGrew, 1997; Flanagan, Ortiz, & Alfonso, 2007) can be used to create theory-driven and comprehensive assessments of cognitive abilities. At the core of this approach is the alignment between the broad abilities (stratum II or III) that one wishes to measure and the abilities that are actually measured by the tests that one wishes to use. This alignment process can identify where there are deficiencies in measuring the desired broad abilities. Additional tests can then be incorporated to ensure that the abilities of interest are adequately measured.
Although the XBA approach has primarily focused on tests more commonly used in non-employment settings (e.g., Wechsler tests), the principles on which the XBA and the CHC theory are based have a number of implications for current practice of using psychometric tests. First, the XBA approach requires that one starts with a theory of intelligence and then aligns the tests to the desired broader abilities from that theory. As has been argued elsewhere (Kaufman, 2000; Scherbaum et al., 2012; Thorndike, 1997), many cognitive ability tests used in personnel assessment are not linked to any theory of intelligence. Moreover, they are not well aligned to the broad abilities that they seek to measure. Consider the distinction described above of organizing broad abilities into those that represent acquired knowledge and those that represent abilities that are independent of a specific domain. A cursory examination of the types of cognitive ability tests that are commonly used in employment contexts would reveal that many of them most closely align with abilities related to acquired knowledge. For example, the Wonderlic Personnel Test consists primarily of items reflecting acquired knowledge in the verbal and quantitative domains. However, many of these tests and the scores obtained from them are described and interpreted as if they assess abilities that are independent of a specific domain. There is often a misalignment between what is measured and the desired broader abilities. This is not to suggest that measuring abilities related to acquired knowledge is unimportant. Such abilities clearly are important, and we would argue that they are potentially becoming increasingly important as the complexity of the workplace grows. However, if one needs an acquired set of knowledge to perform a job, it should be specified and a test of that knowledge should be used rather than testing for general capabilities in a manner that is contaminated with specific acquired knowledge that might not be relevant to the job.
We do suggest that there is an opportunity to better measure intelligence and possibly improve prediction by basing our tests on theory and aligning the measure with the desired construct by following test development principles such as those in the XBA approach. Although these types of suggestions are not new to personnel selection (e.g., Binning & Barrett, 1989), they have yet to substantially impact the development of cognitive ability tests tha...